NewsParser/news_parser.js

// ==UserScript==
// @name         News parser
// @namespace    http://zakonvremeni.ru
// @version      0.1
// @description  Parse news
// @author       AlexeiBv+mirocod@narod.ru
// @match        https://tass.ru/*
// @match        https://ria.ru/*
// @match        https://zakonvremeni.ru/*
// @icon         https://icons.duckduckgo.com/ip2/zakonvremeni.ru.ico
// @grant        none
// ==/UserScript==

// Общественное достояние, 2023, Алексей Безбородов (Alexei Bezborodov) <AlexeiBv+mirocod_platform_bot@narod.ru>

(function() {
    'use strict';

    function getByClass (className, parent) {
        parent || (parent=document);
        var descendants=parent.getElementsByTagName('*'), i=-1, e, result=[];
        var re = new RegExp("(?:^|\\s)" + className + "(?!\\S)");
        while (e=descendants[++i]) {
            if (re.test(e.className)){
                result.push(e);
            }
        }
        return result;
    }

    function GetImageInContainers(baseClass, parent, textAlign) {
        var elems = getByClass(baseClass, parent);
        if (!elems) {
            return '';
        }

        var i;
        var img_src = '';
        var re = new RegExp("(https?:\/\/.*\.(?:png|jpg))");
        for (i in elems) {
            var e = elems[i];
            var children = e.querySelectorAll("*");
            for(let i = 0; i < children.length; i++){
                var c = children[i];
                if (c.nodeName == 'IMG' && re.test(c.src)) {
                    img_src = c.src;
                }
            }
        }
        if (img_src.length > 0) {
            return '<p style = "text-align:' + textAlign + ';"><img src = "'+ img_src + '" width = "600px"/></p>';
        }
        return '';
    }

    function Trim(s) {
        return ( s || '' ).replace( /^\s+|\s+$/g, '' );
    }

    function RemoveBeforeSplitter(a_String, a_Splitter) {
        var index = a_String.indexOf(a_Splitter)
        if (index != -1) {
            return a_String.substring(index + a_Splitter.length);
        }
        return a_String;
    }

    function RemoveAfterSplitter(a_String, a_Splitter, a_SaveSplitter) {
        var index = a_String.indexOf(a_Splitter)
        if (index != -1) {
            var spl_len = a_Splitter.length
            if (!a_SaveSplitter) {
                spl_len = 0;
            }
            return a_String.substring(0, index + spl_len);
        }
        return a_String;
    }

    function FinishWorkFuncTemplate(a_OutTag, a_TextAlign) {
        function FinishWorkFunc(a_Content, a_Element) {
            if (a_OutTag && a_TextAlign) {
                return '<' + a_OutTag + ' style = "text-align:' + a_TextAlign + ';">' + a_Content + '</' + a_OutTag + '>';
                }
            else {
                return a_Content;
            }
        }
        return FinishWorkFunc
    }

    function GetContentInContainers(a_FinishWorkFunc, a_GrubTextFunc, baseClass, parent, a_ElementFilterFunc, a_ClearTextFunc) {
        var elems = getByClass(baseClass, parent);
        if (!elems) {
            return 'Не удалось найти ' + baseClass;
        }

        var result = '';
        for (var i in elems) {
            var e = elems[i];
            if (a_ElementFilterFunc && !a_ElementFilterFunc(e)) {
                continue;
            }

            var content = '';
            if (e.querySelectorAll) {
                var children = e.querySelectorAll("*");
                if (children.length == 0 || e.innerText) {
                    content += a_GrubTextFunc(e);
                }
                else {
                    for (let i = 0; i < children.length; i++) {
                        var c = children[i];
                        content += a_GrubTextFunc(c);
                    }
                }
            }
            if (a_ClearTextFunc) {
                content = a_ClearTextFunc(content);
            }

            if (a_FinishWorkFunc) {
                result += a_FinishWorkFunc(content, e);
            }
            else {
                result += content;
            }
        }
        return result;
    }

    function ClearUrl(a_Url) {
        var separator = '?';
        return RemoveAfterSplitter(a_Url, separator, false);
    }

    function FIlterTrue(element) {
        return true;
    }

    function FIlterRia(element) {
        if (element.dataset.type == 'text' || element.dataset.type == 'quote' || element.dataset.type == 'list') {
            return true;
        }
        return false;
    }

    function FIlterZV(element) {
        if (element.itemprop == 'articleBody') {
            return true;
        }
        return false;
    }

    function ClearTextFuncTemplate(a_RemoveBeforeList) {
        function ClearTextFunc(a_Content) {
            var content = a_Content;
            for (let i = 0; i < a_RemoveBeforeList.length; i++) {
                var r = a_RemoveBeforeList[i];
                content = RemoveBeforeSplitter(content, r);
            }
            return content;
        }
        return ClearTextFunc
    }

    function GrubTextFuncTemplate() {
        function GrubTextFunc(a_Element) {
            var content = '';
            if (a_Element.innerText) {
                    content = Trim(a_Element.textContent);
                }
            return content;
        }
        return GrubTextFunc
    }

    function MakeContent() {
        var content = '';
        var title_tag = 'h2';
        var p_tag = 'p';
        var zero_tag = '';
        var source_add = true;
        var title_func = FinishWorkFuncTemplate(title_tag, 'center')
        var paragraph_func = FinishWorkFuncTemplate(p_tag, 'justify')
        var zero_tag_func = FinishWorkFuncTemplate()
        if (location.hostname == 'tass.ru') {
            content += GetContentInContainers(title_func, GrubTextFuncTemplate(), 'tass_pkg_title--variant_h1_default.*', document.getElementById('content_box'));
            content += GetImageInContainers('Image_wrapper_.*', document.getElementById('content_box'), 'center');
            content += GetContentInContainers(paragraph_func, GrubTextFuncTemplate(), 'Paragraph_paragraph.*', document.getElementById('content_box'), FIlterTrue, ClearTextFuncTemplate(['/ТАСС/. ']));
        }
        else if (location.hostname == 'ria.ru') {
            content += GetContentInContainers(title_func, GrubTextFuncTemplate(), title_tag, document.getElementsByClassName('article__header')[0]);
            content += GetImageInContainers('photoview__open', document.getElementsByClassName('article__header')[0], 'center');
            content += GetContentInContainers(paragraph_func, GrubTextFuncTemplate(), p_tag, 'article__block', document.getElementsByClassName('article__body')[0], 'justify', FIlterRia, ClearTextFuncTemplate(['– РИА Новости. ', '— РИА Новости. ']));
        }
        else if (location.hostname == 'zakonvremeni.ru') {
            var title = GetContentInContainers(zero_tag_func, GrubTextFuncTemplate(), 'page-header', document.getElementsByClassName('item-page')[0]);
            var parent_category = GetContentInContainers(zero_tag_func, GrubTextFuncTemplate(), 'parent-category-name', document.getElementsByClassName('item-page')[0]);
            var category = GetContentInContainers(zero_tag_func, GrubTextFuncTemplate(), 'category-name', document.getElementsByClassName('item-page')[0]);
            var page = RemoveAfterSplitter(Trim(document.getElementsByClassName('item-page')[0].querySelector('[itemprop=articleBody]').textContent), '.', true);
            content = title + '\n' + parent_category + ' ' + category + '\n\n' + page + '\n' + document.URL;
            source_add = false;
        }

        var result = '';
        if (content.length > 0) {
            result = '<textarea id = "news_content" rows="10" cols="100">' + content;
            if (source_add) {
                result += '<p style="text-align: justify;">Источник: <a href = "' + ClearUrl(document.URL) + '">' + location.hostname + '</a></p>';
            }
            result += '</textarea>';
        }
        return result;
    }

    var content = MakeContent();
    var logo = document.createElement("div");
    logo.innerHTML = '<div style="margin: 0pt auto; width: 800px; text-align: center;">' + content + '</div>';

    document.body.insertBefore(logo, document.body.firstChild);

})();